GLM-4.7-Flash GGUF size and VRAM requirements
unsloth/GLM-4.7-Flash-GGUF is a large language model with 29.94 billion parameters, built on the deepseek2 architecture. It is released under the mit license and has been downloaded 94,504 times.
To run unsloth/GLM-4.7-Flash-GGUF locally at a 4,096-token context, its quantized versions need between 8.63 GB (Q1_0, lowest quality) and 56.67 GB (BF16, highest quality) of memory, weights plus KV cache and a system margin included.
For most users the best balance is Q6_K, needing about 23.87 GB. That means unsloth/GLM-4.7-Flash-GGUF fits entirely in the VRAM of a 10 GB GPU or larger, running fully on the GPU.
GGUF file size and memory by quantization
Compare real GGUF weight sizes, estimated KV cache and total memory for Q4, Q5, Q8 and every quantization published in this repository.
| Quant. | Bits | Quality | Weights | KV | Total | Speed~ | Verdict |
|---|---|---|---|---|---|---|---|
| Q1_0 | 2.23 | Very low | 7.76 GB | 0.07 GB | 8.63 GB | 6.4 t/s | Offload |
| IQ1_S | 2.47 | Very low | 8.61 GB | 0.07 GB | 9.49 GB | 5.8 t/s | Offload |
| IQ1_M | 2.62 | Low | 9.13 GB | 0.07 GB | 10.01 GB | 5.5 t/s | Offload |
| IQ2_XXS | 2.81 | Low | 9.79 GB | 0.07 GB | 10.66 GB | 5.1 t/s | Offload |
| IQ2_M | 2.94 | Low | 10.24 GB | 0.07 GB | 11.11 GB | 4.9 t/s | Offload |
| Q2_K | 3.03 | Low | 10.57 GB | 0.07 GB | 11.44 GB | 4.7 t/s | Offload |
| Q2_K_L | 3.05 | Low | 10.63 GB | 0.07 GB | 11.51 GB | 4.7 t/s | Offload |
| Q2_K_XL | 3.18 | Low | 11.07 GB | 0.07 GB | 11.95 GB | 4.5 t/s | Offload |
| IQ3_XXS | 3.45 | Fair | 12.02 GB | 0.07 GB | 12.89 GB | 4.2 t/s | Offload |
| Q3_K_S | 3.55 | Fair | 12.38 GB | 0.07 GB | 13.25 GB | 4.0 t/s | Offload |
| Q3_K_XL | 3.68 | Fair | 12.84 GB | 0.07 GB | 13.71 GB | 3.9 t/s | Offload |
| Q3_K_M | 3.9 | Fair | 13.61 GB | 0.07 GB | 14.48 GB | 3.7 t/s | Offload |
| IQ4_XS | 4.35 | Good | 15.15 GB | 0.07 GB | 16.03 GB | 3.3 t/s | Offload |
| GGUF | 4.53 | Good | 15.8 GB | 0.07 GB | 16.68 GB | 3.2 t/s | Offload |
| IQ4_NL | 4.59 | Good | 15.99 GB | 0.07 GB | 16.86 GB | 3.1 t/s | Offload |
| Q4_0 | 4.6 | Good | 16.03 GB | 0.07 GB | 16.91 GB | 3.1 t/s | Offload |
| Q4_K_S | 4.61 | Good | 16.08 GB | 0.07 GB | 16.96 GB | 3.1 t/s | Offload |
| Q4_K_XL | 4.68 | Good | 16.32 GB | 0.07 GB | 17.19 GB | 3.1 t/s | Offload |
| Q4_K_M | 4.89 | Good | 17.05 GB | 0.07 GB | 17.93 GB | 2.9 t/s | Offload |
| Q4_1 | 5.07 | Very good | 17.67 GB | 0.07 GB | 18.54 GB | 2.8 t/s | Offload |
| Q5_K_S | 5.56 | Very good | 19.39 GB | 0.07 GB | 20.26 GB | 2.6 t/s | Offload |
| Q5_K_M | 5.72 | Very good | 19.94 GB | 0.07 GB | 20.81 GB | 2.5 t/s | Offload |
| Q5_K_XL | 5.8 | Very good | 20.2 GB | 0.07 GB | 21.08 GB | 2.5 t/s | Offload |
| Q6_K | 6.6 | Excellent | 23.0 GB | 0.07 GB | 23.87 GB | 2.2 t/s | Offload |
| Q6_K_XL | 6.99 | Excellent | 24.38 GB | 0.07 GB | 25.25 GB | — | Insufficient |
| Q8_0 | 8.51 | Excellent | 29.66 GB | 0.07 GB | 30.53 GB | — | Insufficient |
| Q8_K_XL | 9.52 | Excellent | 33.18 GB | 0.07 GB | 34.05 GB | — | Insufficient |
| BF16 | 16.01 | Excellent | 55.79 GB | 0.07 GB | 56.67 GB | — | Insufficient |
KV cache computed from the model's exact architecture. Speed is a rough estimate bounded by memory bandwidth.
Frequently asked questions
How much VRAM do you need to run unsloth/GLM-4.7-Flash-GGUF?
You need about 9.49 GB of VRAM to run unsloth/GLM-4.7-Flash-GGUF entirely on the GPU using the IQ1_S quantization (at a 4,096-token context). Smaller quantizations lower the requirement at the cost of quality.
Can I run unsloth/GLM-4.7-Flash-GGUF on an 8 GB GPU?
Partially. unsloth/GLM-4.7-Flash-GGUF only fits on an 8 GB GPU by offloading part of it to system RAM (with Q6_K), which runs but is slower.
Can I run unsloth/GLM-4.7-Flash-GGUF on a 16 GB GPU?
Yes. With 16 GB of VRAM you can run unsloth/GLM-4.7-Flash-GGUF fully on the GPU using Q3_K_M (about 14.48 GB).
Can I run unsloth/GLM-4.7-Flash-GGUF on a 24 GB GPU?
Yes. With 24 GB of VRAM you can run unsloth/GLM-4.7-Flash-GGUF fully on the GPU using Q6_K (about 23.87 GB).
What is the best quantization for unsloth/GLM-4.7-Flash-GGUF?
If memory allows, higher bits-per-weight means better quality. A common sweet spot is a Q4_K_M or Q5_K_M quantization, which keeps most of the quality while roughly halving the memory versus 8-bit. Pick the highest quantization that still fits in your VRAM.